1. Chapter 8 Virtual Memory
Real memory
Main memory, the actual RAM, where a process executes
Virtual memory is a storage allocation scheme in which secondary memory can be addressed as though it were part of main memory
Size is limited by the amount of secondary memory available
Virtual address is the address assigned to a location in virtual memory

2. Keys to Virtual Memory
1) Memory references are logical addresses that are dynamically translated into physical addresses at run time
A process may be swapped in and out of main memory, occupying different regions at different times during execution
2) A process may be broken up into pieces (pages or segments) that do not need to be located contiguously in main memory

3. Breakthrough in Memory Management
If both of those two characteristics are present,
then it is not necessary that all of the pages or all of the segments of a process be in main memory during execution.
If the next instruction and the next data location are in memory then execution can proceed

4. Execution of a Process
OS brings into main memory a few pieces of the program
Resident set: portion of process that is in main memory
Execution proceeds smoothly as long as all memory references are to locations that are in the resident set
An interrupt (memory access fault) is generated when an address is needed that is not in main memory

5. Execution of a Process OS places the process in a blocking state
Piece of process that contains the logical address is brought into main memory
OS issues a disk I/O Read request
Another process is dispatched to run while the disk I/O takes place
An interrupt is issued when disk I/O complete which causes OS to place the affected process in the Ready state

6. Implications of this new strategy
More efficient processor utilization
More processes may be maintained in main memory because only load in some of the pieces of each process
More likely a process will be in the Ready state at any particular time
A process may be larger than main memory
This restriction in programming is lifted
OS automatically loads pieces of a process into main memory as required

7. Thrashing
A condition in which the system spends most of its time swapping pieces rather than executing instructions.
It happens when OS frequently throws out a piece just before it is used
To avoid this, OS tries to guess, based on recent history, which pieces are least likely to be used in the near future.

8. Principle of Locality
Program and data references within a process tend to cluster  only a few pieces of a process will be needed over a short period of time
It is possible to make intelligent guesses about which pieces will be needed in the future, which avoids thrashing
This suggests that virtual memory may work efficiently

9. Performance of Processes in VM Environment
During the lifetime of the process, references are confined to a subset of pages.

10. Support Needed for Virtual Memory
Hardware must support paging and segmentation
OS must be able to manage the movement of pages and/or segments between secondary memory and main memory

11. Paging Each process has its own page table
Each page table entry contains the frame number of the corresponding page in main memory
Two extra bits are needed to indicate:
P(resent): whether the page is in main memory or not
M(odified): whether the contents of the page has been altered since it was last loaded

12. Paging Table
It is not necessary to write an unmodified page out when it comes to time to replace the page in the frame that it currently occupies

13. Address Translation
The frame no. is combined with the offset to produce the real address
The page no. is used to index the page table and look up the frame no.

14. Page Tables Page tables can be very large
Consider a system that supports 231=2Gbytes virtual memory with 29=512-byte pages. The number of entries in a page table can be as many as 222
Most virtual memory schemes store page tables in virtual memory
Page tables are subject to paging

15. Two-Level Hierarchical Page Table
Composed of byte page table entries
Composed of byte page table entries, occupying 210 pages
is composed of kbyte (212) pages

16. Address Translation for Hierarchical page table
The root page always remains in main memory

17. Translation Lookaside Buffer
Each virtual memory reference can cause two physical memory accesses
One to fetch the page table
One to fetch the data
To overcome this problem a high-speed cache is set up for page table entries
Called a Translation Lookaside Buffer (TLB)
Contains page table entries that have been most recently used 17

18. Translation Lookaside Buffer18

19. TLB operation
By the principle of locality, most virtual memory references will be to locations in recently used pages. Therefore, most references will involve page table entries in the cache.
TLB hit
TLB miss 19

20. Page Size Page size is an important hardware design decision
Smaller page size
 less amount of internal fragmentation
 more pages required per process
larger page tables
some portion of page tables must be in virtual memory
double page fault (first to bring in the needed portion of the page table and second to bring in the process page)

21. Page Size Large page size is better because
Secondary memory is designed to efficiently transfer large blocks of data

22. Further complications to Page Size
Small page size
a large number of pages will be available in main memory for a process
as time goes on during execution, the pages in memory will all contain portions of the process near recent references
 low page fault rate

23. Further complications to Page Size
Increased page size
causes pages to contain locations further from any recent reference
the effect of the principle of locality is weakened
 page fault rate rises

24. Example Page Size
The design issue of page size is related to the size of physical main memory and program size.
At the same time that main memory is getting larger, the address space used by applications is also growing.
architectures that support multiple page sizes

25. Segmentation
Segmentation allows the programmer to view memory as consisting of multiple address spaces or segments.
Segments may be of unequal size.
Each process has its own segment table.

26. Segment Table
A bit is needed to determine if segment is already in main memory, if present,
Segment base is the starting address of the corresponding segment in main memory
Length is the length of the segment
Another bit is needed to determine if the segment has been modified since it was loaded in main memory

27. Address Translation in Segmentation
The segment base is added to the offset to produce the real address
The segment no. is used to index into the segment table and look up the segment base

28. Combined Paging and Segmentation
A user’s address space is broken up into a number of segments and each segment is broken into fixed-size pages
From the programmer’s point of view, a logical address still consists of a segment number and a segment offset.
From the system’s point of view, the segment offset is viewed as a page number and page offset

29. Combined Paging and Segmentation
The base now refers to a page table.

30. Address Translation
The frame no. is combined with the offset to produce the real address
The page no. is used to index the page table and look up the frame no.
The segment no. is used to index into the segment table to find the page table for that segment

31. Key Design Elements Fetch policy Placement policy Replacement policy
Cleaning policy
Key aim: Minimise page faults
No definitive best policy 31

32. Fetch Policy Determines when a page should be brought into memory
Demand paging
only brings pages into main memory when a reference is made to a location on the page
many page faults when process first started
Prepaging
pages other than the one demanded by a page fault are brought in.
more efficient to bring in pages that reside contiguously on the disk 32

33. Placement Policy
Determines where in real memory a process piece is to reside
Important in a segmentation system such as best-fit, first-fit, and etc are possible alternatives
Irrelevant for pure paging or combined paging with segmentation 33

34. Replacement Policy
When all of the frames in main memory are occupied and it is necessary to bring in a new page, the replacement policy determines which page currently in memory is to be replaced.
But, which page is replaced? 34

35. Replacement Policy
Page removed should be the page least likely to be referenced in the near future
How is that determined?
Most policies predict the future behavior on the basis of past behavior
Tradeoff: the more sophisticated the replacement policy, the greater the overhead to implement it 35

36. Replacement Policy Frame Locking
If frame is locked, it may not be replaced
Kernel of the operating system
Key control structures
I/O buffers
Associate a lock bit with each frame which may be kept in a frame table as well as being included in the current page table 36

37. Optimal policy
Selects for replacement that page for which the time to the next reference is the longest
Results in the fewest number of page faults but it is impossible to have perfect knowledge of future events
Serves as a standard to judge real-world algorithms 37

38. Optimal Policy Example
The optimal policy produces three page faults after the frame allocation has been filled.

39. Least Recently Used (LRU)
Replaces the page that has not been referenced for the longest time
By the principle of locality, this should be the page least likely to be referenced in the near future
Difficult to implement
One approach is to tag each page with the time of last reference.
This requires a great deal of overhead. 39

40. LRU Example The LRU policy does nearly as well as the optimal policy.
In this example, there are four page faults

41. First-in, first-out (FIFO)
Treats page frames allocated to a process as a circular buffer
Pages are removed in round-robin style
Simplest replacement policy to implement
Page that has been in memory the longest is replaced
But, these pages may be needed again very soon if it hasn’t truly fallen out of use 41

42. FIFO Example The FIFO policy results in six page faults.
Note that LRU recognizes that pages 2 and 5 are referenced more frequently than other pages, whereas FIFO does not. 42

43. Clock Policy Uses an additional bit called a “use bit”
When a page is first loaded in memory or referenced, the use bit is set to 1
When it is time to replace a page, the OS scans the set flipping all 1’s to 0
The first frame encountered with the use bit already set to 0 is replaced.
Similar to FIFO, except that, any frame with a use bit of 1 is passed over by the algorithm. 43

44. Clock Policy Example
incoming page 727 44

45. Clock Policy Example
An asterisk indicates that the corresponding use bit is equal to 1.
The arrow indicates the current position of the pointer.
Note that the clock policy is adept at protecting frames 2 and 5 from replacement. 45

46. Replacement Policy Comparison
Two conflicting constraints:
1. We would like to have a small page fault rate in order to run efficiently,
2. We would like to keep a
small frame allocation. 46

47. Page Buffering
Replacing a modified page is more costly than unmodified page because the former must be written to secondary memory
Solution: page buffering + FIFO
Replaced page remains in memory (only the entry in the page table for this page is removed) and is added to the tail of one of two lists
Free page list if page has not been modified
Modified page list if it has 47

48. Page Buffering
The free page list is a list of page frames available for reading in pages.
When a page is to be read in, the page frame at the head of the list is used, destroying the page that was there.
The important aspect is that the page to be replaced remains in memory.
If the process references that page, it is returned to the resident set of that process at little cost.
In effect, the free and modified page lists act as a cache of pages.

49. Cleaning Policy
A cleaning policy is concerned with determining when a modified page should be written out to secondary memory.
Demand cleaning
A page is written out only when it has been selected for replacement
 Minimizes page writes
 A process that suffers a page fault may have to wait for two page transfers before it can be unblocked  decrease processor utilization 49

50. Cleaning Policy Precleaning
Pages are written out in batches (before they are needed)
 Reduces the number of I/O operations
 Pages written out may have been modified again before they are replaced  waste of I/O operations with unnecessary cleaning operations

51. Cleaning Policy Better approach: incorporates page buffering
Cleans only pages that are replaceable
decouples the cleaning and replacement operations
Pages in the modified list are periodically written out in batches and moved to the free list
 significantly reduces the number of I/O operations and therefore the amount of disk access time
Pages in the free list are either reclaimed if referenced again or lost when its frame is assigned to another page 51